從分配函數構建二維分位向量

Abstract

在本篇論文中，我們以分配函數為基礎，構建兩種分位向量，分別是 NS bivariate quantile 和 The bivariate quantile 。透過模擬，發現當樣本是來自二維對稱型分配時，上述兩種分位向量皆比樣本平均來的有效。

It is well known that the quantile of univariate variable distribution can be constructed in terms of either the empirical distribution function or the associated order statistics. In this paper, we develop two types of quantiles, called the NS bivariate quantile and the bivariate quantile. We also define the bivariate median which is more efficient than the sample mean while the bivariate distribution is symmetric.